Creating Animated 3D Portrait of Red-Haired Girl in Maya & UE5

Introduction

Hello! I'm Andrei. I'm 22 and I've been working in the games industry for 3-4 years now. I started dabbling into this industry by modding the old GTA games at 12 and creating my own skins to make my game look better. Here is my breakdown for my Redhead bust. I hope it will be helpful.

References

I started by gathering an extensive number of unorganized reference images, including real-life models, sculpts, and finished works from artists that I like. In the end, I stuck to a certain grim-looking model and roughly started sculpting to get more of that feel.

Head

If you ever feel like your character's head looks bad, don't worry about it. With a few iterations, you can turn any type of potato into a decent-looking ZBrush sculpt. Just make sure also to get many general artistic anatomy references as well. Pay attention to the planes and curvature of the head by looking from underneath and above.

By the time I went through a few iterations, the head had undergone massive changes in proportion and features, and in the end, I had something looking rather decent (nowhere near finished yet, however). At this point, I decided to wrap a 3D scan store Albedo and the MetaHuman topology on top of it and take it into Unreal Engine to continue the lookdev and the sculpting. I baked the new normal with the combined pores from the scan and imported it into Unreal.

I decided to add makeup early on to get a feel for the final piece and start experimenting with the eye shaders. I also added a hair sculpt and a red slickback. 

So far, I only experimented with the MetaHumans meshes, settings, and tweaking the texture from the attributes within Unreal. Taking it into Substance 3D Painter, I decided to go for a sickly horror feel, completely change the eyebrows shape to a bolder shape, and enhance the Albedo to the maximum with the pore curvature.

At the end of the Substance 3D stage, the head was more or less looking like this:

Eyebrows & Eyelashes

I started grooming the new eyebrows and eyelashes in Houdini and Maya's XGen. It was the first time I ever used Houdini, and I can only attest that it's much more powerful, stable, and procedural, but also a lot harder to power up and use, considering you have to build the entire thing yourself:

For the eyebrows, I recommend using a main collection for the mid-thick part of the eyebrow, with a simple modifier setup made of a big clump, a small clump within the big clump, and a noise modifier. For collection 2, I recommend thinning the strands more and making a nice transition. It's much easier to make a good eyebrow if you split it into multiple collections and have all the control.

For the peach fuzz, make sure to get good references so you see the main body of the peach fuzz and the stray-aways that stick out much longer from time to time in a random manner. I created the guides into Houdini and converted them into curves for Maya. You can also see the direction that those would generally have.

As with the eyebrows, make sure to include two collections, one for the main shorter and more dense body of peach fuzz and another for the longer stray-aways. For the modifiers, use a Cut modifier with a random expression inside. Play with it and always test it inside the engine.

In the end, I was left with a pleasurable result. I also changed the eye shader to something better and started on the hair.

Hair

After the hair was done in XGen, I started playing with the Metapipe plug-in, so I converted my face to the MetaHuman rig. I already had the same topology from the beginning, and I didn't need to rewrap it. The process is straightforward and a bit troublesome to set up and use for the first time, but it gets the job extremely well. I paired the grooms and attached everything, and all went fine. Make sure not to change the scale of the initial MetaHuman body and face, and never make your grooms on a scaled-down model, it will give you problems.

I started experimenting with earrings, accessories, and clothes. I used XGen to make the puff of the cloth of the boust as well. I also made the first lighting setup, which looked decent, but I had to experiment more:

In the end, I came across a much better lighting setup that pushed the model out a lot more. If you're not a lighting artist, experiment and choose from a few lighting iterations. 

In the end, I experimented with a lot of expressions for the presentation and a lot of camera settings as well. This is the final result:

Andrei Dinca, Character Artist

Interview conducted by Gloria Levine